CodepointWidthDetector: reclassify U+25FB, U+25FC as Narrow (#5914)

This seems to be in line with the emoji-sequences table in the latest
version of the Unicode standard: those glyphs require U+FE0F to activate
their emoji presentation. Since we don't support composing U+FE0F, we
should not present them as emoji by default.

Fixes #5910.

Yes, I hate this.

(cherry picked from commit c39f9c6)

src/types/CodepointWidthDetector.cpp

@@ -192,7 +192,7 @@ namespace
         UnicodeRange{ 0x25ce, 0x25d1, CodepointWidth::Ambiguous },
         UnicodeRange{ 0x25e2, 0x25e5, CodepointWidth::Ambiguous },
         UnicodeRange{ 0x25ef, 0x25ef, CodepointWidth::Ambiguous },
-        UnicodeRange{ 0x25fb, 0x25fe, CodepointWidth::Wide }, // OVR 5fb-5fc
+        UnicodeRange{ 0x25fd, 0x25fe, CodepointWidth::Wide },
         UnicodeRange{ 0x2600, 0x2604, CodepointWidth::Wide }, // OVR
         UnicodeRange{ 0x2605, 0x2606, CodepointWidth::Ambiguous },
         UnicodeRange{ 0x2609, 0x2609, CodepointWidth::Ambiguous },